Photo llustration by Thomas Trutschel/Photothek via Getty Images
Researchers at Duo Security have developed an algorithm to hunt Twitter bots at Twitter-sized scale.
Why it matters: On a big social media platform, using humans to hunt automated scam accounts is a particularly difficult game of whack-a-mole. That is exactly the kind of problem Duo's algorithm can help solve. One of the moles it whacked during a testing was a large network of cryptocurrency scammers.
Researchers Olabode Anise and Jordan Wright will present that research later this week at the Black Hat cybersecurity conference. Twitter was responsive when Duo's reported bots, they said.
The big picture: The sheer scope of the bot problem makes it difficult to fight — just ask Elon Musk, who had so many bots using his name to promote cryptocurrency scams that Twitter banned naming accounts after him.
- The bots in question are unmanned accounts that automatically spew content onto Twitter.
- Not all bots are malicious. Most aren't. "During testing, one of the things we discovered was a bot that posted song lyrics someone thought were interesting," Anise told Axios.
- The algorithm does not separate good from evil, just human from machine. Real humans still need to make the final call.
How it works: Duo feeds publicly available data, like how quickly accounts reply to other accounts, screen names, the time of day of posting and data from profiles.